Methods

Study design and population

In this single-center retrospective cohort study, regular-care data were extracted from medical records of people with bleeding disorders who underwent TKA or THA who attended the Van Creveldkliniek at University Medical Center Utrecht between 1989 and 2025. Patients with moderate or severe hemophilia A or B (FVIII/FIX <5%) or von Willebrand disease (VWD) with a FVIII activity <5% were included, as arthropathy in patients with these factor levels is assumed to be bleeding-related [35]. All types of primary THA or TKA were included. All patients were followed according to standard care. Patients could be included multiple times with distinct joints. The study was approved by the institutional Research Quality Coordinator of Division Internal Medicine and Dermatology of the University Medical Center Utrecht (25U-0426). Informed consent was waived because the health care data were anonymized. Patients that objected to the use of their (anonymized) health care data were excluded.

Standard care

The Van Creveldkliniek at UMC Utrecht is a multidisciplinary hemophilia treatment center providing lifelong care, with regular monitoring and treatment of bleeding disorders and close surveillance of joint health through routine visits and imaging. Physiotherapy was used to prevent or treat limitations in physical functioning as a consequence of HA and to rehabilitate after joint bleeds. When persistent pain and functional limitations remain, TJR is indicated. For every surgery, the local hemostatic protocol was used: factor levels were increased to 80% to 100% at day of surgery and at day 1. From days 2 to 7, factor levels are kept at 50%, and from days 7 to 10 (THA) or 14 (TKA), factor levels are kept at 30% either by continuous infusion of coagulation factor or bolus therapy. Because surgeries are complex and expensive in this population, and multiple joints might be affected, procedures performed on different joints might be combined in one admission (multiple joint procedures), such as bilateral procedures or together with an ankle arthrodesis. In uncomplicated, unilateral procedures, (partial) weightbearing is allowed from day 0. In case of revision surgery, poor joint quality, or other complications, the surgeon determines the weightbearing protocol. In multiple-joint procedures involving ankle arthrodesis, 6 weeks of cast immobilization is required, during which no weightbearing was permitted. In case of catheter angiography, this is performed by interventional radiologists prior to TKA to evaluate the presence and extent of synovial hyperemia in the knee. Selective genicular artery embolization is performed in case of synovial hyperemia to reduce hyperemia to limit the chance of postoperative joint bleeding [36]. The life cycle of the prosthesis is monitored biannually after 10 years and annually after 15 years.

Primary outcome

The primary outcome was time to prosthesis failure, measured in years. Prosthesis failure was defined as revision surgery or a documented indication for revision in patients who were ineligible for surgery or declined operative treatment. Time to failure was calculated as the interval between the date of surgery and the date of prosthesis failure. Patients who died during follow-up were censored at the date of death.

Patient characteristics

The following patient characteristics were extracted from the electronic patient records: age at time of surgery (years); type and severity of bleeding disorder; type of surgery (TKA/THA); year of surgery; body mass index (BMI) at time of surgery; treatment regimen at time of surgery (on demand/factor prophylaxis/nonfactor prophylaxis); HIV, hepatitis C, and inhibitor (defined as anti-FVIII above 1.0 or 0.4 for surgeries before and after 2005, respectively) status at time of surgery (positive/negative); multiple joint procedure (yes/no); and degree of arthropathy (defined as total Pettersson score before the first TJR; if unavailable, the score closest to the first TJR date if within 3 years). The Pettersson score rates the 6 joints most prone to HA (elbow, knee, ankle) on a 13-point scale [37]. Higher scores indicate more severe arthropathy.

Perioperative characteristics

The following perioperative data were extracted: difference in pre- and postoperative hemoglobin value (defined as: last known preoperative value minus the first value at day 1 after surgery), length of hospital stay (days), preoperative angiography (yes/no, TKA only), cementation of femoral stem (yes/no, THA only), and postoperative weightbearing protocol (full/limited weightbearing).

Long-term complications

The following data on long-term complications were extracted: postoperative bleeding (yes/no; defined as: hemarthrosis, wound dehiscence, or signs of internal bleeding within 1 month after surgery), PJI (yes/no), and loosening or liner wear diagnosed by the orthopedic surgeon (yes/no) at any time during follow-up.

Analysis

Patient and perioperative characteristics and long-term complications were reported as medians with ranges or frequencies with percentages. Time to prosthesis failure of primary TKA and THA was used as a dependent variable in the Kaplan–Meier survival analyses. PSRs and complication rates were derived in percentages with CIs. Kaplan–Meier results were interpreted until the number of patients at risk (nR) dropped below 10 [38]. Univariable Cox proportional hazards (PH) regression analyses were used to identify predictors for failure. Age, severity of disorder (dichotomized as severe hemophilia A/B or nonsevere and VWD), treatment regimen (dichotomized as on-demand or prophylactic), year of surgery, BMI, HIV status, hepatitis C status, inhibitor status, preoperative angiography (only for TKA), degree of arthropathy, multiple joint procedure, hemoglobin difference, length of hospital stay, preoperative angiography, cementation (only for THA), weightbearing protocol, PJI, and postoperative bleeding were used as independent variables in univariable analysis. Variables with a univariable association P < .1 were included in a multivariable model. If multivariable models included postoperative variables, a post hoc analysis including only preoperative variables were conducted to increase the clinical applicability of the results. Fisher’s exact test was used to test differences between complication rates before and after 2000. The year 2000 was chosen as it was used in the most relevant meta-analysis and marked important advances in hemophilia care [8]. Missing data was handled by multiple imputation with predictive mean matching with 5 imputations. Results were pooled afterward. A sensitivity analysis was performed to correct for multiple episodes of TJR in the same individuals at different time points; the same analyses were repeated including only the patient’s first surgery. For patients whose first surgery was bilateral, every right side was included. Significance was set at P < .05. All analyses were performed in RStudio (2025) with use of packages “survival,” “mice,” “icenReg,” and “survminer”.